Chapter 4: Network Interface Cards
After reading the chapter and completing the exercises, the students should be able to:
• Describe what role a network adapter card plays in networked communications
• Explain how network adapters prepare data from transmission, accept incoming network traffic, and
control how networked communications flow
• Understand the variety of configurable options for network adapters and describe common settings
• Describe important characteristics for selecting adapter cards
• Recount network adapter enhancements that can improve performance
• Explain the role of driver software in network adapters
Network Interface Card (NIC) Basics
1. Note that the sole purpose of a network interface card is to mediate the connection between the
computer and the network medium. This includes managing the physical connection as well as
translating data from the computer into signals appropriate for the network medium.
2. Note that network interface cards go by many names including network adapter, NIC, and just plain
From Parallel to Serial, and Vice Versa
1. Describe parallel transmission. Data is sent along a number of lines at the same time. This is the
transmission method used within the computer.
2. Describe serial transmission. Data is sent in a linear sequence. This is the transmission method used by
most network media. One of the jobs of the NIC is to convert the data it receives in parallel to serial for
transmission across the network, and perform the reverse operation for incoming data. Figure 4-1
illustrates this function very well.
3. Define bus width. This is the number of parallel lines that make up the computer’s bus. This width is
measured in the number of bits, which can be moved in a single bus transfer. Although buses are
discussed in greater detail in the next section, mention here that the first PCs used an 8-bit bus. The
second generation of PCs moves to 16-bit, which became the Industry Standard Architecture (ISA).
4. Describe the role of the transceiver in network communication. As the students learned in Chapter 3,
the transceiver for a Thickwire network is connected directly to the backbone. The computer is then
attached to the transceiver via a drop cable. However, most other types of NICs include a transceiver
directly on the card. It is also very common to include multiple transceivers on the NIC to
accommodate different media.
5. Note that the NIC is responsible for packaging data into packets, which are the fundamental units of
data for network communication. The NIC is also responsible for correcting packet-level errors.
6. Describe the NIC’s role as “gatekeeper” for the computer. It examines all packets on the network as
they are sent and determines whether the computer is the destination of any packet. It does this by
examining the destination network address of the packet and comparing it to its own network address.
This process is discussed in more detail in Chapter 5.
1. Note that there are five types of PC busses: Industry Standard Architecture (ISA), Enhanced Industry
Standard Architecture (EISA), Micro Channel Architecture (MCA), Peripheral Component Interface
(PCI), and Accelerated Graphics Port (AGP).
2. Describe each of the busses and their speeds (8-bit, 16-bit, or 32-bit). Note that EISA slots can
accommodate ISA cards. Also note that MCA is not often found in PCs.
Other PC Interfaces Used for Networking
1. Discuss how USB and FireWire interfaces are used in networking. USB is a relatively low-speed serial
interface that is very popular for connecting peripheral devices like mice, keyboards, printers,
scanners, and cameras. FireWire is a high-speed serial bus that operates at up to 3200 Mbps and is
governed by the IEEE 1394 and 1394b standards.
Principles of NIC Configuration
1. Mention Plug and Play and its ideal configuration strategy. Also mention, however, that it often does
not work. The more devices in your computer, the less likely it is that Plug and Play will work.
2. Note that many cards are software configurable, but some cards, especially older cards, require setting
be made via jumpers or DIP switches.
1. Define interrupt and interrupt request line (IRQ). These are used by the devices in a computer to get
the attention of the CPU.
2. It is very important the students understand that each device must have its own IRQ. If two devices are
configured to use the same IRQ, problems occur. Depending on the conflict, the machine may not even
3. Note that diagnostic software such as MSD or WINMSD can be used to view the interrupts being used
in a computer. If you run MSD before you install the NIC, you will know which IRQs are available.
4. For purposes of the Microsoft certification test, it is important the students know which IRQs are
assigned by default to which devices. There may be questions, which say “Your computer has a floppy
drive, two serial ports and one printer port. Which IRQ can be used for a network adapter?”
Base I/O Ports
1. Define I/O ports. Again, it is important for the students to know which I/O ports are available for use
by peripherals such as NICs.
Base Memory Address (membase)
1. Define base memory address and how it is used by the NIC to communicate with the CPU.
2. Also describe the high memory area. This goes back to the earliest days of the PC when, according to
Bill Gates in a speech in 1981, “640K ought to be enough for anybody.”
Making the Network Attachment
1. Note that the network card must match the medium. Some cards, like Ethernet, are able to
accommodate a number of media, whereas the more advanced and specialized cards are only able to
support one media.
1. What type of data communication takes place between the network interface card and the
Answer: Parallel transmission
2. Which PC bus architecture is used only for video adapters?
Answer: Accelerated Graphics Port (AGP)
3. What computing technique is used by peripherals to request CPU time?
Answer: Interrupt requests via interrupt request lines (IRQs)
4. What is the default IRQ for the keyboard? The floppy disk controller? The parallel port? The primary
hard drive controller?
Answer: 1, 6, 7, 14
Choosing Network Adapters for Best Performance
1. Discuss each of the methods listed for increasing the speed of adapter cards. There are a number of
questions on the Microsoft certification test, which require the students to know how server speed can
be increased. The questions are phrased “Which of the following network adapter card configurations
will provide the highest throughput? ISA with DMA, EISA with shared system memory, PCI with
DMA, or MCA with bus mastering?”
1. As mentioned in Chapter 3, network adapters for wireless communication are different than those for
wired communication. Note that these are available for most operating systems.
Remote Boot Adapters
1. Discuss the need for specialized network adapters to be used in diskless computers. Often, for security
reasons, computers are installed without disk drives of any nature. The network adapters in these
computers use a Boot PROM to connect to the network to load the operating system.
1. Discuss the role of the device driver in network communications. This driver is used by the computer
to control the NIC.
2. Note that in the early days of networking, a driver was compiled for each variation of the network card.
For example, a driver was created for a card using IRQ 5 and I/O port 300h. Then, if the card was
move to another machine or reconfigured for any reason, a new driver was compiled with the new
settings. As you can image, this was an exceptionally tedious task.
3. Discuss the NDIS, WDM, and ODI standards, which allow more than one protocol to run on a NIC
and, in the case of ODI, simplify driver development. It is important for the students to be familiar
with these standards for the Microsoft test.
1. What NIC feature is concerned with traffic management?
Answer: Quality of Service (QoS)
2. Which computers use remote boot adapters and Boot PROMs?
Answer: Diskless workstations
3. Which driver software standard defines a complete driver, rather than specific portions of the OSI
Answer: Win32 Driver Model (WDM)
Additional Case Projects
Develop an example of how bus width effects communication.
There are many different things the students could use for this example. The key is that when there are
more lines to access a component, communication is faster. Some examples might include waiting in line at
the grocery store or a concert.
Your company often employs temporary workers to input sensitive data. As your network and database
have grown, concern has mounted that the information these workers enter may not be safe, or may be
taken off the premises. Your boss has asked you to implement a hardware-based solution to ensure that
nothing happens to the data in the system.
The best option for this scenario is to install remote boot adapters in diskless workstations. This way, the
workers are able to input the data, but are not able to damage the data with viruses or remove information
on diskette. It is important that the students understand that changes will need to be made on the
workstation side and also on the server to support remote boot systems.
Internet resources in the book:
Additional Internet resources:
ZDNet’s ZDWebopedia is an extensive site that defines and explains computer and networking terms.
Check the definitions for chapter subjects like IRQs, ISA, and DIP switches. ZDWebopedia is at:
Technical Notes for the Hands-On Projects
The lab setup for the Chapter 4 projects includes the following elements (see the table):
Network adapter and driver software
HANDS-ON NETWORK WORKSTATION OTHER
PROJECT DEVICES OR SERVER RESOURCES REQUIRED
4-1: Document your PCs Standard workstation
4-2: Install a network Network adapter Standard workstation
4-3: Check the hardware Standard workstation
4-4: Install the driver Network adapter Standard workstation
software driver software
4-5: Test the installation Standard workstation
4-6: Find information on Internet access Standard workstation
Chapter Four Solutions
Answers to Chapter Review Questions
1 d. IRQ 11
2 a. PCI, b. MCA, c. EISA
3 A; as for b., a driver may act like part of the operating system, but is really a separate
piece of software
4 a. agree with
5 False; it’s the other way around
6 d. 16-bit and 32-bit
7 d. on-board RAM
8 All answers (a., b., c., d.)
9 All answers (a., b., c., d., e.)
11 c. Neither device will work.
12 d. Permit the CPU and a peripheral device to exchange messages.
13 All answers (a., b., c., d.)
14 b. EISA, c. MCA, d. PCI
15 c. Windows 95, e. Windows 2000 Professional
16 c. To inform the CPU that some event requiring its attention has occurred.
17 a. ISA, b. EISA
18 All answers (a., b., c., d.)
19 b. IRQ, c. base I/O port, d. base memory address. The correct answers do not include e.
since selecting a transceiver is rare. It only occurs on combo cards with multiple
connections. Most NICs only have a single connection type.
20 b. transceiver
21 All answers (a., b., c., d., e.)
22 a. Networks got faster, so buses did; too, b. Increased CPU speeds demanded faster,
wider busses; c. Increased application sophistication increased user’s demands for data;
d. More applications needed network support.
23 a. Check the CPU, motherboard, and peripherals at PC boot time.
24 a. Macintosh OS 8.0 and higher, e. Windows NT Server and Workstation
25 c. 280h, d. 300h
27 b. WDM
28 c. Boot PROM
29 d. FireWire
Hands-on Projects Discussion
The Hands-on projects for this chapter involve installing a network interface card in a computer. The
projects assume the students will be using Windows machines. It is important during the second project that
students ground themselves properly and work with the equipment very carefully.
Hands-on Project 4-1
Learning how to find, investigate, and document your computer’s configuration is an important task that
any system administrator must understand. Although methods vary from Windows version to Windows
version, it’s vital to learn how to elicit this information from all of these versions. Likewise, mapping out a
system’s configuration is not only a valuable tool whenever you must add or remove hardware from that
system, it also helps to maintain an accessible source of information whenever troubleshooting or repair is
Hands-on Project 4-2 through 4-4
Learning how to install a network adapter, or other PC adapter card, is an important part of system
administration; whether it is to set up new systems for initial use or to upgrade or maintain such systems for
ongoing use. Try to remember that it pays to take your time, to avoid brute force, and to treat the equipment
with care and respect. It is possible to ruin cards or bus slots if too much force is applied; it’s better to take
your time and ask for help if you get stuck or find yourself in an awkward situation. Likewise, learning how
to install and configure adapter drivers, and to install and configure necessary network protocols, is also a
key administrative task. Although you can’t do permanent damage by making mistakes during these phases
of the process, you can easily render a machine unable to access the network. Since that’s the whole point
of the installation, learning the steps (and the troubleshooting skills that come from making and fixing
mistakes) will provide invaluable experience in the workplace.
Hands-on Project 4-5
Checking your work is particularly important when adding or removing hardware from a PC. Not only
should you make sure the machine powers up correctly, but it’s also smart to exercise the capabilities that a
new adapter card provides, as well as trying the system’s other capabilities as well. Running a performance
testing utility like the one included with Norton Utilities checks your system thoroughly and will let you
know if it finds anything amiss (also note that Norton’s System Information utility makes an excellent
alternative to the built-in Windows utilities which, though they do the job, do not do as nice a job of
bringing all configuration information together into a single display or printable report).
Hands-on Project 4-6
The only drivers that Microsoft can include on the software distributed with an operating system are those
that already exist when the operating system release is put together, or generic drivers that offer limited
performance and functionality. Given the importance of good network performance, and the odds that
newer PCs will also contain newer network interfaces, it’s highly likely that you will have to find or update
driver software for your NICs at some point in your networking career. That’s why this exercise shows you
how to visit a vendor’s Web site, and locate and download a specific driver. Although you will probably
visit multiple such Web sites, this exercise shows you the basic steps involved in obtaining the software
Case Projects Discussion
The hardware that's installed in the PC includes a serial mouse and an internal modem; this does away with
both serial ports, so there go IRQs 3 and 4. The two SCSI controllers will probably take up IRQs 10 and 11.
Also, IRQs 0 (system time), 1 (keyboard), 2 (IRQ cascade), 6 (floppy drive), 7 (LPT1:), 8 (real-time clock),
9 (IRQ cascade), 13 (math coprocessor), and 14 (IDE controller) are often taken. This leaves only IRQ 5
available from the list shown, which is why it's also the most common factory default setting for NICs
Because a 32-bit bus mastering NIC with additional on-board RAM gives you about as much speed as you
can get when attaching a server to a cable segment, it definitely produces the required result of handling
network traffic with reasonable response time. This also helps to guarantee that the server won't slow the
network down. But since it would be cheaper (in fact half the cost) to use plain-vanilla NICs on the
relatively idle segments, this solution fails to keep the hardware costs to a minimum. Thus, the correct
answer is b, which produces the required result but only one of the optional desired results.
The real issue with Bob's PC is that it is painfully slow. This will affect the network only when he's
transmitting data, however. Therefore, replacing his NE1000 with an NE2000 is probably the most
practical solution (answer a). Replacing the card with an EISA version does no good because an EISA card
won't work in an ISA slot (although ISA cards can work in EISA slots, it doesn't work the other way
around). If Bob generates a lot of traffic on the network, however slowly, answer c will indeed isolate other
users from his influence; in that case, this answer makes the most sense. Since you can't replace Bob's old
PC with a new one, a PCI card won't work in his machine, either, so answer d is completely incorrect.
Because a is more practical than c, we think the correct answer to this question should be a but it's a close
Dealing with interrupt conflicts gets to be a way of life when installing NICs, especially for heavily-loaded
end user machines. The only real solution to the problem that doesn't diminish the machine's current
capabilities is to follow the strategy outlined in answer a, and try to change the IRQ for some other device
to free up an IRQ that the NIC can use. Answer b implies that whatever devices can't be made to work can
be omitted, which the user may not appreciate. Answer c may work, but the PC Card NIC will still need an
IRQ, too—you may be spending more money, just to wind up with the same problem you were trying to
solve. Answer d limits the speed of the network attachment to serial port speeds; unless you spend real
money ($150 or more) for a high-speed serial interface, you'll wind up with a network attachment that runs
significantly slower than a NIC attachment. This is a solution that's only worth considering if all other
options have been exhausted and the machine still can't be connected to the network.
Key Terms for Chapter Four
adapter slot — The sockets built into a PC motherboard that are designed to seat adapter cards. See also
Accelerated Graphics Port, Industry Standard Architecture, Extended Industry Standard Architecture,
Micro Channel Architecture, and Peripheral Component Interface. (All represent specific types of adapter
AGP (Accelerated Graphics Port) — A special-purpose bus used solely to interconnect PCs with a
graphics adapter and one or more display devices. AGP is a high-speed, 64-bit-wide bus capable of
bandwidth from 0.25 to 1.0 Gbps.
base I/O port — The memory address where the CPU and an adapter check for messages that they leave
for each other.
base memory address — The memory address at which the transfer area between the computer’s main
memory and a NIC’s buffers begin, bounded by the size of its extent. See extent.
Boot PROM — A special programmable chip that includes enough software to permit a computer to boot
sufficiently and access the network. From there, it can download an operating system to finish the boot
boot up — The process a computer goes through when starting.
buffer — A temporary storage area that a device uses to contain incoming data before it can be processed
for input or to contain outgoing data before it can be sent as output.
bus — Also called the bus architecture, a specialized collection of parallel lines in a PC used to ship data
between the CPU and peripheral devices and, occasionally, from one peripheral device to another. One or
both adapters involved must have bus-mastering capabilities.
bus mastering — The quality of an adapter card’s circuitry that allows it to take possession of a
computer’s bus and coordinate data transfers without requiring any service from the computer’s CPU.
bus width — The number of parallel lines that make up a particular kind of computer bus. For example, ISA
supports 8- and 16-bit bus widths, EISA and MCA support 16- and 32-bit bus widths, and PCI supports 32- and
64-bit bus widths.
device driver — A software program that mediates communication between an operating system and a
specific device for the purpose of sending and/or receiving input and output from that device.
diagnostic software — Specialized programs that can probe and monitor a system (or a specific system
component) to determine if it works properly and, if not, to try to establish the cause of the problem.
DIP switch — An electrical circuit that consists of a series of individual two-way switches contained in a
single chip. See also dual inline package (DIP), which explains the pin-outs for this kind of package.
Direct Memory Access (DMA) — A technique for addressing memory on some other device as if it were
local memory directly available to the device accessing that memory. This technique lets a CPU gain
immediate access to the buffers on any NIC that supports DMA.
diskless workstations — Network computers that require a special type of ROM because they have no
built-in hard or floppy drives.
DMA — See Direct Memory Access.
driver — An abbreviation for “device driver,” a small program that mediates between an operating system
and the hardware device it knows how to access.
dual inline package (DIP) — An integrated computer circuit that features two parallel rows of pins of
equal length, offset approximately 1 cm.
EISA — See Extended Industry Standard Architecture.
Extended Industry Standard Architecture (EISA) — A 32-bit PC bus architecture that is backward-
compatible with the older, slower 16-bit ISA bus architecture.
extent — The size of an area; usually describes the upper limit of a memory region on a PC named by a
base address that indicates the starting point (upper bound = base address + extent).
fault tolerance — A feature of a system, which allows it to continue working after an unexpected
hardware or software failure.
FireWire — A high-speed, external serial bus that supports bandwidths up to 400 Mbps and can connect up to 63
devices. Also known as IEEE 1394. FireWire is used for streaming video and multimedia, and networking, and to
attach video devices to computers.
Hardware Compatibility List (HCL) — A vendor-maintained list of all hardware compatible with a
particular operating system. In practice, it is the name of a document maintained by Microsoft that lists all
hardware compatible with Windows NT or Windows 2000.
HCL — See Hardware Compatibility List.
hexadecimal — A mathematical notation for representing numbers in base 16. The numbers 10 through 15
are expressed as A through F; 10h or 0¥10 (both notations indicate the number is hexadecimal) equals 16.
high memory area (HMA) — The region of memory on a PC between 640 KB and 1024 KB (usually
referred to in hex as A0000 through 100000). This is the area where device driver buffer space and shared
system memory are typically allocated.
HMA — See high memory area.
Industry Standard Architecture (ISA) — Originally an 8-bit PC bus architecture but upgraded to 16-bit
with the introduction of the IBM PC/AT in 1984.
Interrupt Request Line (IRQ) — Any of 16 unique signal lines between the CPU and the bus slots on a
PC. IRQs define the mechanism whereby a peripheral device of any kind, including a network adapter, can
stake a claim on the PC’s attention. Such a claim is called an “interrupt,” which gives the name to the lines
that carry this information.
IPSec (IP Security)—An Internet security protocol that’s gaining acceptance as a way to protect network
traffic from unwanted snooping.
IRQ — See Interrupt Request Line.
ISA — See Industry Standard Architecture.
jumper — A small, special-purpose connector designed to make contact between two pins on an adapter
card of some kind; sometimes used to establish configuration settings on network cards and other computer
jumper block — A collection of two or more sets of jumper pins or a special connector designed to make
contact between two or more sets of contiguous jumper pins at the same time.
MAC — See Media Access Control.
MAC address — The address on any NIC that the MAC layer handles.
MCA — See Micro Channel Architecture.
Media Access Control (MAC) — A level of data communication where the network interface can directly
address the networking media; also refers to a unique address programmed into network adapters to
identify them on any network where they might appear.
membase — An abbreviation for base memory address; see base memory address.
Micro Channel Architecture (MCA) — IBM’s proprietary 16- and 32-bit computer buses. Originally
developed for its PS/2 PCs, MCA is now popular on its midrange RISC/6000 computers.
MSD.EXE — The Microsoft diagnostics program that ships with DOS, Windows 3.x, and Windows 95
operating systems. This program can document IRQs, base memory addresses, and HMA regions in use.
NDIS — See Network Device Interface Specification.
network adapter — A synonym for network interface card (NIC). It refers to the hardware device that
mediates communication between a computer and one or more types of networking media. See also
network interface card.
network address — The number that identifies the physical address of a computer on a network. This
address is hard-wired into the computer’s NIC.
network card — Synonym for network interface card.
Network Device Interface Specification (NDIS) — A standard for providing an interface between a
network interface card and the network medium that enables a NIC to use multiple protocols.
network interface card (NIC) — The hardware device that mediates communication between a computer
and one or more types of networking media.
NIC — See network interface card.
ODI — See Open Data-link Interface.
on-board co-processor — A special- or general-purpose microprocessor that appears on an adapter card,
usually to offload data from a computer’s CPU. NICs with on-board co-processors usually employ the
Open Data-link Interface (ODI) — A specification developed by Apple Computer and Novell that
simplified driver development and enabled the use of multiple protocols from a single NIC.
packet — A specially organized and formatted collection of data destined for network transmission;
alternatively, the form in which network transmissions are received following conversion into digital form.
parallel transmission — The technique of spreading individual bits of data across multiple, parallel data
lines to transmit them simultaneously, rather than according to an ordinal and temporal sequence.
PCI — See Peripheral Component Interface.
Peripheral Component Interface (PCI) — The 32- and 64-bit PC bus architecture that currently prevails
as the best and fastest of all available bus types.
Plug and Play — The Microsoft requirements for PC motherboards, buses, adapter cards, and operating systems,
which let a PC detect and configure hardware on a system automatically. For Plug and Play to work properly, all
system components must conform rigorously to its specifications; currently, only Windows 95, Windows 98, and
Windows 2000 support this architecture.
POST — See power-on self-test.
power-on self-test (POST) — The set of internal diagnostic and status-checking routines a PC and its
peripheral devices perform each time the computer is powered on.
Quality of Service (QoS) — A networking term that specifies a guaranteed level of service when applied
to applications that require high bandwidth.
RAM buffering — A memory-access technique that permits an adapter to use a computer’s main memory
as if it were local buffer space.
security feature — In terms of NICs, a feature that allows the card to handle security-related protocols,
including encryption services.
serial transmission — A technique for transmitting data signals, which set each bit’s worth of data (or its
analog equivalent) one at a time, one after another, in sequence.
shared adapter memory — A technique for a computer’s CPU to address memory on an adapter as if it
were the computer’s own main memory.
shared system memory — A technique for an adapter to address a computer’s main memory as if it were
resident on the adapter itself.
traffic management — Also called traffic grooming. In terms of NICs, features that improve network
accessibility for remote users, especially those using applications that require higher bandwidth, such as
streaming video or multimedia.
transceiver — A device that transmits and receives network information.
Universal Serial Bus (USB) — A hot-pluggable Plug and Play serial interface that operates at a maximum
data transfer rate of 12 Mbps. USB ports support peripheral devices such as mice, keyboards, and other
pointing devices in addition to some printers, scanners, telephony equipment, and monitors.
USB — See Universal Serial Bus.
Win32 Driver Model (WDM) — A unified driver architecture that allows a single
driver to be written for both Windows 95 and Windows NT.
WINMSD.EXE — The Windows NT 4.0 built-in diagnostics program. WINMSD.EXE can report on
IRQs, base memory addresses, HMA use, and other internal system data.